Research map / Vision-language and multimodal / CLIP and zero-shot
DroneFINE: Domain-Aware Parameter-Efficient Fine-Tuning of Vision-Language Detectors for Drone Images
Wu Ke, Yanan Zhang, Yingjie Gao, Wenhao Li, Chenyu Zhou, Xinzhu Ma, Jiaxin Chen, Di Huang
Read the paperCheck your idea against 137,217 papers
AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| add foreground awareness to HyperAdapter (ablation) | HyperAdapter · VisDrone | mAP50: 60.2 -> 61.1 | Improved |
| add foreground awareness to HyperAdapter (ablation) | HyperAdapter · UAVDT | mAP: 24.9 -> 26.1 | Improved |
| add contrastive loss to SemanticGate (ablation) | SemanticGate · VisDrone | mAP50: 60.4 -> 60.8 | Improved |
| add contrastive loss to SemanticGate (ablation) | SemanticGate · UAVDT | mAP50: 42.1 -> 42.0 | No clear effect |
| combine HyperAdapter and SemanticGate into DroneFINE-T instead of Mona + CoOp (ablation) | Mona + CoOp · VisDrone | mAP: 37.5 -> 38.4 | Improved |
| add HyperAdapter to baseline (ablation) | Baseline · VisDrone | mAP50: 59.6 -> 61.1 | Improved |
| add ObjSeeker foreground module to HyperAdapter (ablation) | HyperAdapter · VisDrone | mAP: 38.1 -> 38.0 | No clear effect |
| use HyperAdapter instead of Adapter + ObjSeeker (ablation) | Adapter + ObjSeeker · VisDrone | mAP50: 60.0 -> 61.1 | Improved |
| increase rank from 64 to 96 (ablation) | HyperAdapter · VisDrone | mAP: 38.4 -> 37.9 | Hurt |
| lower rank from 64 to 32 (ablation) | HyperAdapter · VisDrone | mAP: 38.4 -> 37.3 | Hurt |
| use meaningful background vocabulary instead of Random-20 words (ablation) | SemanticGate · VisDrone | mAP50: 59.4 -> 60.8 | Improved |
| increase random vocabulary size from 5 to 20 words (ablation) | SemanticGate · VisDrone | mAP50: 59.6 -> 59.4 | No clear effect |
Closest papers by meaning
- Multimodal Causal Reasoning for UAV Object Detection (NeurIPS 2025 · similarity 0.87)
- RemDet: Rethinking Efficient Model Design for UAV Object Detection (AAAI 2025 · similarity 0.86)
- PointSR: Self-Regularized Point Supervision for Drone-View Object Detection (CVPR 2025 · similarity 0.85)
- Self-Prompting Analogical Reasoning for UAV Object Detection (AAAI 2025 · similarity 0.85)
- Roboflow100-VL: A Multi-Domain Object Detection Benchmark for Vision-Language Models (NeurIPS 2025 · similarity 0.84)
- Multiview Aerial Visual RECognition (MAVREC): Can Multi-view Improve Aerial Visual Perception? (CVPR 2024 · similarity 0.84)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.
